White — Specific Heats of Silicates and Platinum. 341 



mean heat, the result is /' (0) or the true heat. Hence if the 

 mean heats are plotted and a curve drawn through them (fig. 2) 



Fig. 2.' 



Temperature. 

 Fig. 2. Diagram, illustrating graphic determination of true specific heat. 



and the tangent to any point n of the curve is produced to the 

 y axis, say at q, the projection, p q, of the tangent on this axis, 

 added to the mean heat, gives the true heat for the tempera- 

 ture of n. 



Experimental results. — Most of the results so far obtained 

 are given in Tables 1 and 2, and graphically in figures 3 and 4. 

 The values for orthoclase and the glass of the correspond- 

 ing composition* were obtained with the wide furnace first 

 used. Determinations of wollastonite and pseudo-wollastonite 

 with this furnace ran about one per cent above those obtained 

 later, hence the orthoclase and "orthoclase glass" results have 

 been arbitrarily lowered two-thirds of a per cent. The platinum 

 determinations were also made in this earlier furnace, but have 

 not been altered, as the good thermal conductivity of the 

 platinum probably reduced the systematic error considerably. 

 Of course, these results are not quite as certain as the later 

 ones. Otherwise, the results from 700° to 1300° probably 

 contain no error exceeding 0*5 per cent. This is, of course, 

 merely given as the result of careful consideration of all 

 sources of error. Obviously, positive statements as to system- 



* Made by melting the orthoclase, which does not crystallize on cooling. 

 See Day and Allen, Publication No. 31, Carnegie Institution of Washington, 

 p. 50, 1904 ; this Journal, (4), xix, 120, 1905. 



Am. Jour. Sci.— Fourth Series, Vol. XXVIII, No. 166.— October, 1909. 



